I worked today on a formalisation of the co-referenced actions coding schema. I used the three levels model introduced yesterday to mark how such actions are generated. I made some simple assumption on how to disambiguate between similar cases and how to order them:
 Actions weight more than words. In other words, commitment is expressed by actions more than with words. Agreeing at verbal level does not imply agreement at knowledge level.
 Actions planned and executed by different agents weights more than the same actions planned and performed by the same agent.
 Grounding at knowledge level would require completing the three phases: Plan, Commitment, Feedback.